Auto¶

library(ggplot2)

library(RColorBrewer)

pacman::p_load(plotly, tidyverse, reshape2)

library(IRdisplay)

# save default options and parameters

defop = options()

defpar = par(no.readonly=T)

​

# function to modify plot parameters

plot_pars = function(w=7,h=7) {options(repr.plot.width=w, repr.plot.height=h)}

# Import data

fdir = r"(E:\Data Science\Statistics\Intro to Statistical Learning with R)"

fpath = file.path(fdir,'datasets','Auto.csv')

auto = read.csv(fpath)

​

dim(auto)

head(auto)

# check for missing values

anyNA(auto)

# structure of dataset

str(auto)

# Rows with non-numeric values

auto[which(is.na(as.numeric(auto$horsepower))),]

# is.numeric(auto$horsepower) will not give boolean array but a single True/False value.

# Rows with ? in horsepower

which(auto$horsepower == '?')

# Rows with ? in any column

auto[which(apply(auto, 1, function(x) any(x %in% c("?")))), ]

# Deleting rows with ?

auto = auto[-which(auto$horsepower=='?'), ]

sum(auto=="?")

​

dim(auto)

class(auto$horsepower)

# Convert horsepower to numeric

auto$horsepower = as.numeric(auto$horsepower)

str(auto)

# No. of unique values in columns

sapply(auto, function(x) length(unique(x)))

# Quantitative predictors

quant_data = auto[, setdiff(names(auto), c('cylinders','origin','name'))]

names(quant_data)

summ = data.frame(sapply(quant_data, range), row.names=c('min','max'))

summ['range',] = summ['max',] - summ['min',]

summ

summ['mean',] = sapply(quant_data, mean)

summ['sd',] = sapply(quant_data, sd)

round(summ, 2)

# Removing rows 10-85

df1 = quant_data[-seq(10,85), ]

dim(df1)

# Check if the 10th row in filtered dataset matches 86th row of unfiltered dataset

all(df1[10,] == quant_data[86, ])

summ = data.frame(sapply(df1, range), row.names=c('min','max'))

summ['range',] = summ['max',] - summ['min',]

summ['mean',] = sapply(df1, mean)

summ['sd',] = sapply(df1, sd)

round(summ, 3)

# Check sum

sapply(df1, sum)

# Pairs plot panels

panel.hist <- function(x, col="thistle", ...)

{

    usr <- par("usr"); on.exit(par(usr))

    par(usr = c(usr[1:2], 0, 1.5) )

    h <- hist(x, plot = FALSE)

    breaks <- h$breaks; nB <- length(breaks)

    y <- h$counts; y <- y/max(y)

    rect(breaks[-nB], 0, breaks[-1], y, col = 'whitesmoke', ...)

}

panel.cor <- function(x, y, digits = 2, prefix = "", cex.cor, ...)

{

    usr <- par("usr"); on.exit(par(usr))

    par(usr = c(0, 1, 0, 1))

    r <- abs(cor(x, y))

    txt <- format(c(r, 0.123456789), digits = digits)[1]

    txt <- paste0(prefix, txt)

    if(missing(cex.cor)) cex.cor <- 0.8/strwidth(txt)

    text(0.5, 0.5, txt, cex = cex.cor * r)

}

# Hue - cylinder

col_map = setNames(c('#F1DCD7','#E3BAC0','#CD9BB1','#AE80A3','#61566E'), sort(unique(auto$cylinders)))

cols = col_map[unlist(as.factor(auto$cylinders))]

​

plot_pars(17,15)

pairs(quant_data, cex=2.5, cex.labels = 2, cex.axis=1.6, pch=21, bg=cols, col='whitesmoke', lwd=0.3,

     diag.panel=panel.hist, lower.panel=panel.cor, oma=c(3,3,15,3))

par(xpd = TRUE)

legend(x=0.25, y=1, fill=unlist(col_map), legend=names(col_map), horiz=T, bty="n")

options(defop)

# Scatter plot - origin as hue

fig = plot_ly(auto, x=~year, y=~mpg, color=~as.factor(origin), width=700, height=400,

              colors = brewer.pal(length(unique(auto$origin)),"Set1"),

              symbol=~origin, symbols = c('triangle-up','circle','x'),

              text = ~paste('mpg:',mpg, '<br>name:',name, '<br>origin:',origin, '<br>cyl:',cylinder),

              type="scatter", mode='markers', marker=list(size=8)) %>%

      hide_colorbar() %>%

      layout(showlegend = TRUE, legend = list(orientation="h", xanchor="center", x=0.45, y=1),

             hoverlabel=list(bgcolor="white"), title = "Scatter Plot",

             xaxis = list(title = "Year", zeroline=F, showgrid=F),

             yaxis = list(title = "mpg", zeroline=F, gridcolor="white")) %>%

      config(displayModeBar = FALSE)

​

options(jupyter.plot_mimetypes=c("text/html","image/svg+xml"))

display(fig)

​

# https://rstudio-pubs-static.s3.amazonaws.com/448200_6bb02977b4c04e0da508ac0131f71d48.html

# https://colorbrewer2.org/#type=qualitative&scheme=Set1&n=3

# https://plotly.com/python/marker-style/#custom-marker-symbols

# symbol=~origin, symbols = c('square','circle','x','+','x','triangle')

# use symbol('^') to get list and names of acceptable symbols

# config(displayModeBar = FALSE) >> hides plotly bar  (https://plotly-r.com/control-modebar.html)

# css colors >> https://www.w3.org/TR/css-color-3/#svg-color

# hover format >> https://plotly.com/r/hover-text-and-formatting/

# color=~as.factor(origin) shows the legend, then hide_colorbar() not reqd

# Color map

cols = character(nrow(auto))

cols[] = '#61566E'

cols[auto$origin == 1] = '#EDD1CB'

cols[auto$origin == 2] = '#AD6E94'

​

# Hue - origin

plot_pars(15,15)

pairs(quant_data, cex=2.2, cex.labels = 2, cex.axis=1.6, bg=cols, pch=21, col='whitesmoke',

     diag.panel=panel.hist, lower.panel=panel.cor, oma=c(3,3,16,3))

par(xpd = TRUE)

legend(x=0.4, y=1,, fill=c('#EDD1CB','#AD6E94','#61566E'),

       legend=c(levels(as.factor(auto$origin))), horiz=T, bty="n")

options(defop)

# Frequency distribution of cylinders

table(auto$cylinders)

# year-wise distribution of cars with cylinder count

cyl_year = auto %>% count(year,cylinders) %>% spread(cylinders, n, fill=0)

cyl_year

# year-wise distribution of cars with cylinder count

df <- melt(cyl_year ,  id.vars='year', variable.name='cylinders')

​

p = ggplot(data=df, aes(x=year, y=value)) + 

        geom_line(linetype='twodash', aes(color=cylinders)) + 

        geom_point(aes(color=cylinders)) + 

        theme_classic() + 

        scale_color_brewer(palette="Paired") + 

        theme(legend.title=element_text(size=10), legend.text=element_text(size=8))

​

fig = ggplotly(p, width=650, height=350)

fig = fig %>% config(displayModeBar=T) %>% 

layout(fig, plot_bgcolor='white', 

       xaxis=list(showticklabels=T, tickmode="linear", tickformat='0', dtick=1, 

                  tickfont=list(size=10), zeroline=F, showgrid=F, color="black",

                  title=list(text="year", standoff=10, font=list(size=15))),

       yaxis=list(showticklabels=T, tickmode="linear", tickformat='0', dtick=5, 

                  title=list(text="mpg", standoff=15, font=list(size=15)),

                  gridcolor="white", showticklabels=T, tickfont=list(size=10)))

​

options(jupyter.plot_mimetypes=c("text/html","image/svg+xml"))

display(fig)

# setup 1 - classic ggplot theme

df <- melt(cyl_year ,  id.vars='year', variable.name='cylinders')

plot_pars(7,3)

fig = ggplot(data=df, aes(x=year, y=value)) + 

    geom_line(linetype='twodash', aes(color=cylinders), size=0.8) + 

    geom_point(aes(color=cylinders), size=2) + 

    theme_classic() + 

    scale_color_manual(values=c('#fdc086','#386cb0','#beaed4','#33a02c','#f0027f')) + 

    scale_x_continuous(breaks = seq(min(df$year), max(df$year), by=1)) + 

    scale_y_continuous(breaks = round(seq(min(df$value), max(df$value), by=5), 1)) + 

    theme(text=element_text(size=13), panel.background=element_rect(fill="white"),

          panel.grid.major.x=element_blank(), panel.grid.major.y=element_blank(),

          panel.grid.minor.x=element_blank(), panel.grid.minor.y=element_blank(),

          axis.text.x=element_text(color="grey20", size=9, angle=0, hjust=.5, vjust=.5, face="plain"),

          axis.text.y=element_text(color="grey20", size=9, angle=0, hjust=1, vjust=0, face="plain"),  

          axis.title.x=element_text(color="grey20", size=12, angle=0, hjust=.5, vjust=0, face="plain"),

          axis.title.y=element_text(color="grey20", size=12, angle=0, hjust=.5, vjust=.5, face="plain"))

fig